Abstract Age-related macular degeneration (AMD) causes vision loss in millions worldwide. Central to AMD initiation and progression is the retinal pigment epithelium (RPE), which is clinically visualized via the massed autofluorescence (AF) of its lipofuscin and melanolipofuscin granules. We hypothesize, based on our imaging and pathology studies, that AMD can be staged and monitored by the expression and distribution of unique RPE fluorophores. We identified and characterized ex vivo spectral signatures of distinct fluorophore families in normal RPE, Bruch's membrane and drusen, AMD's hallmark lesion. However, the molecular sources of these spectra in the macula are now uncertain due to seminal imaging mass spectrometry (IMS) studies showing that a major lipofuscin fluorophore, A2E, is abundant in the periphery. Thus, the long-term goal of this 6-investigator collaboration is to develop AMD diagnostics based on hyperspectral AF for spectral, molecular biopsy of the RPE, linking clinical pathology to underlying molecular composition. Hyperspectral AF imaging, unlike conventional AF imaging, acquires 3-dimensional ?hypercubes? of data (2 spatial coordinates ? x, y - and 1 spectral - wavelength). We explored imaging data with novel tensor-based tools exploiting multiple excitation wavelengths to discover RPE spectral signatures and their spatial distributions. We propose to link fluorophores to granules, RPE cells, tissue, and AMD stages in 3 aims using a common human tissue source. Aim 1 uses hyperspectral AF tissue mapping to understand AMD pathology at the spectral level in eyes with AMD and unaffected control eyes. We will map RPE flat-mounts by hyperspectral AF microscopy linked to a pathology grading system. Spectral AF components will be recovered mathematically and assigned to subcellular and extracellular features. Aim 2 will quantify AMD pathology at the subcellular level by enumerating fluorophore-containing granules using structured illumination microscopy and 3-dimensional electron microscopy. Aim 3 uses hyperspectral fluorophore identification to understand AMD pathology at the molecular level. We will determine candidate molecules for the major spectral components discovered in Aim 1 by hyperspectral thin layer chromatography and will verify their spatial distributions by IMS. Synthetic authentic standards will ensure spectral validation. Biological validation at this level is unprecedented for clinical ophthalmology, yet warranted by the size of the AMD patient population, the enormity of knowledge gap about major RPE fluorophores in human eyes, the availability of donor tissue, and the proven success of validating other imaging technologies. Our results will directly translate to clinical hyperspectral AF imaging for noninvasive, spatially precise early detection and longitudinal AMD follow-up, in vivo target discovery, and immediate extensions beyond AMD. From our discoveries will flow a huge range of experiments in outer retinal cell biology to deepen understanding of retinal degenerations.